Contrastive Learning for Cross-modal Artist Retrieval

Music retrieval and recommendation applications often rely on content features encoded as embeddings, which provide vector representations of items in a music dataset. Numerous complementary embeddings can be derived from processing items originally represented in several modalities, e.g., audio signals, user interaction data, or editorial data. However, data of any given modality might not be available for all items in any music dataset. In this work, we propose a method based on contrastive learning to combine embeddings from multiple modalities and explore the impact of the presence or absence of embeddings from diverse modalities in an artist similarity task. Experiments on two datasets suggest that our contrastive method outperforms single-modality embeddings and baseline algorithms for combining modalities, both in terms of artist retrieval accuracy and coverage. Improvements with respect to other methods are particularly significant for less popular query artists. We demonstrate our method successfully combines complementary information from diverse modalities, and is more robust to missing modality data (i.e., it better handles the retrieval of artists with different modality embeddings than the query artist's).